{"id":"W4289657037","doi":"10.2196/35150","title":"Assessing the Generalizability of Deep Learning Models Trained on Standardized and Nonstandardized Images and Their Performance Against Teledermatologists: Retrospective Comparative Study","year":2022,"lang":"en","type":"article","venue":"JMIR Dermatology","topic":"Cutaneous Melanoma Detection and Management","field":"Medicine","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"National Health and Medical Research Council; Medical Research Council; Australian Government; Nvidia","keywords":"Generalizability theory; Artificial intelligence; Computer science; Machine learning; Psychology; Medical physics; Medicine; Developmental psychology","routes":{"ca_aff":false,"ca_fund":false,"ca_venue":true,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0004836751,0.0002179949,0.0007571398,0.0001476417,0.0006937223,0.00004430415,0.00008893417,0.00004711296,0.00003336059],"category_scores_gemma":[0.00004591401,0.0001481006,0.00007038446,0.0002121688,0.0003852248,0.0001008751,0.0001952114,0.0004859596,3.778142e-7],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001469442,"about_ca_system_score_gemma":0.00004586065,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.000008363744,"about_ca_topic_score_gemma":0.000006762014,"domain_scores_codex":[0.9979396,0.0007855708,0.0003965286,0.0003779934,0.0002678819,0.0002323746],"domain_scores_gemma":[0.9991098,0.0002210026,0.0002259794,0.0002655061,0.0001162449,0.00006146888],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"qualitative","study_design_scores_codex":[0.01261768,0.003991272,0.781785,0.00107675,0.002877872,0.001260196,0.11495,0.01201733,0.01398895,0.002611701,0.001233241,0.05159003],"study_design_scores_gemma":[0.02992993,0.00650364,0.3244962,0.0001066756,0.0004538276,0.01037833,0.366659,0.2446923,0.008989115,0.0009747855,0.005725272,0.001090941],"study_design_candidate":"observational","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9927049,0.0002636164,0.001425844,0.0006869033,0.00007686498,0.001551985,0.00001510621,0.00008460753,0.003190122],"genre_scores_gemma":[0.9992319,0.00006356016,0.0001235035,0.000257187,0.00001137838,0.0002486357,0.00000921297,0.00001547308,0.00003909815],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4572888,"threshold_uncertainty_score":0.6039367,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03285068772704493,"score_gpt":0.314296387602292,"score_spread":0.2814456998752471,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}